Development and Validation of an Optical SPR Biosensor Assay for

In recent years there has been an increase in the use of tylosin in apiculture as bacterial brood diseases become resistant to oxytetracycline. Confir...
2 downloads 0 Views 91KB Size
J. Agric. Food Chem. 2005, 53, 7367−7370

7367

Development and Validation of an Optical SPR Biosensor Assay for Tylosin Residues in Honey MARIANNE CALDOW,*,† SARA L. STEAD,† JOANNA DAY,† MATTHEW SHARMAN,† CHEN SITU,‡ AND CHRIS ELLIOTT‡,§ Central Science Laboratory, Sand Hutton, York YO41 1LZ, United Kingdom, Department of Veterinary Science, Queen’s University Belfast, Stormont, Belfast BT4 3SD, United Kingdom, and Veterinary Sciences Division, Department of Agriculture and Rural Development, Stormont, Belfast BT4 3SD, United Kingdom

In recent years there has been an increase in the use of tylosin in apiculture as bacterial brood diseases become resistant to oxytetracycline. Confirmatory mass spectrometry based methods have been developed but up until now there has been no complementary screening method available capable of sub 10 µg kg-1 detection limits. In this paper the development and validation of a screening method using optical biosensor technology is presented. The honey was first dissolved in a phosphate buffer and following solid-phase extraction (SPE) cleanup was analyzed using a Biacore Q instrument. Using the criteria specified in European Commission Decision 2002/657/EC for qualitative screening methods, the detection capability (CCβ) of the method was determined to be 2.5 µg kg-1. Honey samples containing trace residue levels of tylosin were analyzed by both the biosensor screening method and a LC-MS/MS confirmatory procedure; the results were in good agreement. KEYWORDS: Tylosin; biosensor; honey; veterinary drug residues; food safety

INTRODUCTION

Tylosin is a macrolide bacteriostatic antibiotic (made naturally by the bacterium Streptomyces fradiae), active against most Gram-positive bacteria, mycoplasma, and certain Gram-negative bacteria. The mode of action of tylosin is via inhibition of bacterial protein synthesis by selective binding to the 50S ribosome, a cellular structure only present in some bacteria (1). The macrolide class of antibiotics, including tylosin, have been widely used as therapeutic agents and growth-promoting antibiotics in veterinary medicine. However, the use of tylosin as a feed additive was banned by the European Union in 1999 (2). The chemical structure of tylosin A (the major bioactive component) is shown in Figure 1. Oxytetracycline (OTC) is commonly used worldwide and is the only legally approved antibiotic registered in the USA for the control of the bacterial pathogens, American Foulbrood (AFB) (Paenibacillus larVae White) and European Foulbrood (EFB) (Melissococcus plutonius) within bee colonies (3). The emergence of bacterial resistance to oxytetracycline is now widespread within the United States and oxytetracycline is no longer a fully effective treatment for the control of AFB. Therefore, new antibiotics have been investigated for the control * To whom correspondence should be addressed. Fax: +44 1904 462311. Tel.: +44 1904 462000. E-mail: [email protected]. † Central Science Laboratory. ‡ Department of Veterinary Science, Queen’s University Belfast. § Department of Agriculture and Rural Development.

Figure 1. Structure of tylosin A.

of AFB (3). The most active antibiotics which are used in the control of OTC-resistant AFB are erythromycin, lincomycin, monensin, and tylosin (4). The usage trends of these other antibiotics in apiculture appear to be linked to geographical origin, which can be correlated to the spread of bacterial resistance to OTC. It has been reported that tylosin can be administered by veterinarians as an “ExtraLabel Use Privilege” (when suffering or death of the animal may result from failure to treat) (5). Tylosin residues in honey in the range 0.0012-0.1156 mg kg-1 have recently been reported by the Canadian Food Inspection Agency (CFIA) (6). EU Maximum Residue Limits (MRLs) exist for tylosin in tissues of all food-producing species; however, there is currently no MRL set for tylosin in honey (7). Therefore, there must be no detectable residues of tylosin in honey resulting from its “Extra-Label” or unauthorized use (5). Within this laboratory sub 10 µg kg-1 tylosin residues have been detected in retail honey samples as part of the Department for Environment Food and Rural Affairs (Defra) funded UK Non-Statutory Surveillance scheme. With the possible increase

10.1021/jf050725s CCC: $30.25 © 2005 American Chemical Society Published on Web 08/19/2005

7368

J. Agric. Food Chem., Vol. 53, No. 19, 2005

Caldow et al.

in unauthorized usage of antibiotics in apiculture, it has therefore become essential to develop rapid and sensitive analytical methodology to monitor for potential residues of tylosin in honey samples. Work has been conducted and previously reported by both Wang (8) and Thompson et al. (9, 10) concerning the development of LC-MS/MS methodology to quantify and confirm tylosin residues in honey with associated low-level detection limits. At present, there are no reports in scientific literature of a screening method capable of detecting tylosin residues in honey at sub 10 µg kg-1 to complement the detection limits achievable by LC-MS/MS. Charm Sciences Inc. markets a semiquantitative test for tylosin in honey using the Charm II 1600 scintillation analyzer; however, the detection limit reported is 15-20 µg kg-1 (11). The aim of this investigation was to develop and validate a sensitive complementary screening method for the analysis of tylosin using SPR biosensor technology. MATERIALS AND METHODS Apparatus and Reagents. The SPR-based biosensor system Biacore Q was obtained from Biacore AB (Uppsala, Sweden). Antibody and immobilized sensor chips were provided by Queen’s University Belfast. HBS-EP buffer (HBS) was purchased from Biacore AB. Biacore control software, version 3.0.1, was used for instrument operation. Standards were purchased from Sigma (Poole, UK). Strata-X 33 µm Polymeric sorbent SPE cartridges were purchased from Phenomenex (Macclesfield, UK). Potassium dihydrogen orthophosphate, sodium hydroxide, HPLCgrade water, methanol, and acetonitrile were purchased from Fisher Scientific (Loughborough, UK). Honey. Yorkshire Blossom Honey (both runny and set), known to be free from tylosin, was obtained from the UK National Bee Unit and was used as blank controls throughout the method development and validation experiments. Retail honey from a variety of geographical locations and floral origins were purchased for use within the matrix specificity experiment. Honey samples from retail surveillance programs known to contain residues of tylosin were also included in the experiments. Antibody Production. Protein conjugates of tylosin-HSA were prepared and mixed with Freund’s adjuvant and used as immunogens for raising polyclonal antibodies in New Zealand White rabbits by the method of McCaughey et al. (12). The first injections of animals were performed using Freund’s complete adjuvant which contained heatkilled Mycobacterium tuberculosis. Subsequent injections were performed using Freund’s incomplete adjuvant. The emulsions containing 1 mg of immunogen were injected subcutaneously into four sites of the animal (left and right front quarters and left and right hind quarters). The animals received immunization every 2 weeks and a blood sample was collected at the same time as the injection assessment of antibody activity with a competitive ELISA using drug-HRP as label. When a suitable titer was achieved, antiserum was harvested from the rabbit and stored at -20 °C. The specificity of the polyclonal antibody was assessed by cross-reactivity studies with a comprehensive range of antimicrobial agents (13). Immobilization of the Sensor Chip. Tylosin was immobilized onto the surface of a CM5 sensor chip. The chip surface was activated with 40 µL of a 1:1 mixture of 0.4 mol dm-3 1-ethyl-3(3-dimethylaminopropyl)-carbodiimide hydrochloride (EDC):0.1 M N-hydroxysuccinimide (NHS) for 20 min followed by 50 µL of 10 mmol dm-3 hydrazine hydrate for 1.5 h at room temperature. The unreacted sites were blocked by the addition of 50 µL of 1 mol dm-3 ethanolamine for 20 min. The reactants were removed and 50 µL of tylosin (2 mg mL-1 in 10 mmol dm-3 sodium carbonate/bicarbonate buffer, pH 9.6) was added and allowed to remain in contact with the sensor chip surface overnight at room temperature. The reactant was removed and the chip surface was washed with deionized water, then dried using a stream of nitrogen gas, and stored desiccated at +4 °C when not in use. Extraction Procedure. (Based on the method of extraction previously reported by Wang (8)). Aliquots of honey (5 g) were dissolved

Figure 2. SPR biosensor calibration curve using matrix-matched standards in the range 0.05−10 µg kg-1. in 20 mL of 0.1 mol dm-3 potassium buffer (pH adjusted to 8). Nonpolar interferences (such as waxes) were partitioned into hexane (10 mL). An aliquot (15 mL) of the aqueous fraction was collected and further cleaned up using a Strata-X solid-phase extraction (SPE) cartridge; the cartridge was preconditioned with methanol (10 mL), water (10 mL), and extraction buffer (5 mL). The extract was loaded at a flow rate of 1 mL min-1; the loaded cartridge was first washed with 40% methanol in water (5 mL) and eluted with methanol (5 mL). Following evaporation (at 40-50 °C) the residue was reconstituted in HBS-EP buffer (500 µL), prior to analysis using the Biacore Q. For comparative quantitation by LC-MS the residue was reconstituted in methanol (500 µL). Biosensor Assay. The extracts were mixed within the autosampler with an equal volume of stock antibody diluted in running buffer (1 in 300) and injected for 180 s over the sensor chip surface at a flow rate of 40 µL min-1. The sensor surface was regenerated between cycles with a 90 s pulse of 20% acetonitrile in 0.25 mol dm-3 sodium hydroxide at a flow rate of 40 µL min-1. Calibration. For each batch of samples a set of five matrix-matched standards at 0.5, 1.0, 2.0, 5.0, and 10 µg kg-1 were prepared by fortifying blank extract. The use of matrix-matched standards compensated for any nonspecific binding effects. Duplicate injections at each concentration were performed and the mean response was used to construct a calibration curve. A typical calibration curve is given in Figure 2. The working range (80 to 20% inhibition) of the sigmoidal curve is between 0.5 and 4.0 µg kg-1. Batch recovery was determined by the fortification of honey samples prior to extraction. The concentration was determined against the matrixmatched calibration curve. Percentage recovery was calculated accordingly (observed concentration/added concentration) × 100. LC-MS/MS Determination. (Based on the LC-MS/MS conditions previously reported Wang (8)). The LC-MS/MS system used was a Quattro Ultima Platinum Triple Quadrupole coupled to an Alliance 2695 Separations Module (Waters). The software used was Mass Lynx version 4.0. A gradient separation was performed on a Thermo Electron Corporation HyPurity C18 5 µm particle size, 150 × 2.1 mm with a C18 guard column installed. The transitions monitored were 917>174, 917>773, and 917>156. Method Validation. The validation of the SPR biosensor method and determination of CCβ were conducted in accordance with the European Commission Decision, 2002/657/EC (14); details of those performance criteria required for screening analysis can be found in Table 9 of the document. Both analyte and matrix specificity experiments were conducted to determine the method cross-reactivity and its applicability to a wider variety of honey types from different geographical locations. Honey, known to contain residues of tylosin, has been analyzed using both biosensor and LC-MS/MS quantification and the results compared.

Optical SPR Biosensor Assay for Tylosin Residues in Honey

J. Agric. Food Chem., Vol. 53, No. 19, 2005

7369

Table 1. Comparative Analysis of Different Honey Samples by SPR Biosensor and LC-MS/MS concentration (µg kg-1)

Figure 3. Structure of spiramycin. RESULTS AND DISCUSSION

Method Validation. Determination of CCβ. Since no MRL currently exists for tylosin in honey, an initial assessment of method performance was made via the calibration graph (Figure 2). Whilst this data indicated that the developed method had a “limit of detection” of approximately 0.5 µg kg-1, a formal experiment was required to calculate the “detection capability” (CCβ) of the biosensor screening method (14). To calculate the CCβ, 20 samples of known honey blank were spiked with tylosin at 2.5 µg kg-1. This concentration was chosen to minimize the occurrence of false positives. During initial method validation some honey varieties, when tested, elicited a low-level biosensor response which would be equivalent to tylosin in the range 0.3-1.98 µg kg-1. However, these potential residues did not confirm by LC-MS/MS. All recoveries were in the range 47.6-70.4% with an average of 60.3% and a % CV of 9.1. From these results it can be concluded that the CCβ for this screening procedure is less than or equal to 2.5 µg kg-1. Intra-laboratory Repeatability. An intra-laboratory repeatability experiment was conducted by a second analyst. Replicate samples (n ) 7) were fortified with tylosin at 2.5 µg kg-1 (equivalent to the CCβ). The calculated recoveries were in the range 54.8-70.0% with an average of 61.5% and a % CV of 7.5, which compared well to the primary validation results. Ruggedness. A series of key parameters such as the capacity of the SPE cartridges and the pH and molarity of the extraction/ loading buffer were investigated. SPE cartridges with sorbent bed capacity of both 200 and 60 mg were tested with no significant difference noted in the results obtained. Variations in the extraction/loading buffer specifications were investigated with each of the following tried: water (pH unadjusted), water at pH 8, 0.1 mol dm-3 buffer (native pH approximately 4.4) unadjusted in pH, buffer adjusted to pH 8 with (10% variation in molarity (0.08, 0.09, 0.10, 0.11, and 0.12 mol dm-3). These experiments were not exhaustive but indicated that variations in those conditions tested did not significantly affect the performance of the method. Analyte Specificity. The specificity of the entire procedure was investigated against some structurally related compounds and compounds which may be administered concurrently: tilmicosin, spiramycin, erythromycin, bacitracin, and lincomycin. To determine the cross-reactivity of the five compounds, blank honey samples were fortified with each of the different compounds in the range 0.5-10 µg kg-1. The IC50 values were calculated and expressed as a percentage relative to the IC50 value for tylosin. As expected, tilmicosin, erthromycin, bacitracin, and lincomycin did not show any measurable crossreactivity. Spiramycin (Figure 3) produced a measurable response on the biosensor and its cross-reactivity was found to be approximately 60%.

country of origin/honey variety

LC-MS/MS

SPR biosensor

Scottish Blossom Yorkshire Spring Blossom Hungarian Acacia New Zealand Clover French/Spanish Lavender European Sunflower Scottish Lowlands Spanish Rosemary Spanish Eucalyptus Spanish honeycomb (Balearic) - 1 Blenda Orange Blossom Mexican Wildflower Turkish Pine Forest Tasmanian Leatherwood English Dorset Heathland Australian Eucalyptus Scottish Heather Australia/New Zealand Manuka Clear Greek Spanish honeycomb (Balearic) - 2 USA Clear Pure Naturalb Hungarian Acacia Italian Chestnut